Have Financial Development and Private Capital Inflows
Served as Influential Determinants of Economic Growth in
Central Europe? An Empirical Study of the Visegrad Four
Countries
Ricardo Jimenez 10041745 August 22nd, 2013
Supervisor: Konstantinos Mavromatis University of Amsterdam, MSc Economics
1
1
INTRODUCTION ... 2
2
BACKGROUND ... 4
2.1 Transition Economies of Central Europe ... 4
2.2 The European Union and Central Europe ... 5
2.3 Recent Economic Trends in Central Europe ... 7
2.3.1 Capital Inflows ... 7
2.3.2 Economic Growth in Central Europe... 8
2.3.3 Financial Crisis in Central Europe ... 9
3
THEORY AND LITERATURE REVIEW ... 10
3.1 Economic Growth ... 10
3.2 Private Capital Inflows ... 13
3.3 Financial Development ... 15
4.
EMPIRICAL ANALYSIS ... 23
4.1 Data ... 23 4.1.1 Country Sample ... 23 4.1.2 Data Collection ... 24 4.1.3 Data Variables ... 24 4.1.4 Regression Model ... 25 4.1.5 Data Summary ... 26 4.2 Empirical Results ... 274.3 Long-Run Relationships: Unit Root and Cointegration Testing ... 28
4.4 Discussion ... 33
5
CONCLUSION ... 35
REFERENCES ... 37
APPENDIX 1 - TABLES ... 40
2 1 INTRODUCTION
The EU accession group1 of 2004 consisted of countries with recent transition economies, deriving primarily from Central Europe and the Baltic area. In the previous decade prior to EU membership, this group endured vast market and institutional transformations in preparation of strengthening their economies. Consequently, these countries experienced relatively higher economic growth compared to the more established Western European economies, both pre and post 2004 accession period. Established economic growth theory indicates that among others, education, quality of institutions, the rule of law and openness to trade are to account for this output expansion. More importantly, having sound
macroeconomic indicators, such as GDP, inflation, unemployment and international trade, are crucial for growth, and they are even better suited variables to explicitly measure economic conditions. This thesis will empirically test if two additional factors, the inflow of private capital from abroad (FDI) and the process of financial development, were key contributing factors to such escalated economic growth.
Both factors have potentially large exogenous effects which can positively (or
negatively) impact economic growth. Private capital inflows promote business activity and job creation, which can be measured, so their impact on growth is more transparent. The second factor, financial development, brings about exogenous benefits such as improved efficiency via better resource allocation (investment funding), which is more complicated to measure and account for its impact on economic growth. However, financial
development is also a long process that brings forth sophisticated domestic banking and stock markets, albeit potentially unwarranted risk as they become more integrated with world markets. For example, the recent financial crisis which originated in 2007 gave way to an unquestionable global dip in world output. However, countries in Central Europe showed signs of quicker recovery, while richer and more developed ones struggled to recover at the same speed. Since the financial crisis, by name, roots from the
interconnectedness of developed financial systems, then one can assume that the further financial development had evolved in Central Europe, the more it would struggle to recover as well. In fact, Central European countries are intertwined with the financial systems of
1
3 well-developed EU members, as it is an implicit element for economic cooperation. It can then be inferred that Central Europe should have experienced similar volatile events as those witnessed in more developed Western EU nations, but were not present.
The question is then: Is there a strong link between both private capital inflows and financial development and the observed period of high economic growth in this region? The position of this paper is that foreign direct investment, a main component of private capital flows, has positively impacted economic growth, while financial development has had no impact, due to its early stages of evolvement. The analysis will focus on a group of countries in Central Europe known as the Visegrad Four (V4 henceforth), and includes the Czech Republic, Hungary, Poland, and Slovakia. Aside for sharing similar socio-cultural histories, these four countries present an interesting case study for several reasons. First, in 2008, the V4 accounted for 84%2 of total GDP for the Central European accession group of 2004. Additionally, the V4 constitutes 86%3 of the population of the aforementioned accession group, indirectly indicating the economic influence this group can have in Central Europe. Second, the V4 shares a beneficial commonality, which is having Germany as their largest export market. Again, this indirectly signals a potential benefit due to Germany’s economic prowess not only within the EU, but also in the world. Third, all four constituent countries of the V4 are veteran members4 of the Organization for Economic Cooperation and Development (OECD). Only two other nations of the accession group of 2004 recently share this privilege5. Entering the OECD requires both a certain level of economic
development, as well as a series of examinations to assess a country’s ability to meet OECD standards in a wide range of policy areas6. Therefore, such status reinforces the second reason of the group’s potential leadership role within Central Europe.
In this paper, the impact of investment capital flows and financial development on economic growth will be empirically tested with a panel data regression for the period 1993 to 2011. The purpose is not to determine if proper policies have been implemented to foster financial development, but rather to see how far along these countries are in the
2
Source: World Bank Database
3
Source: World Bank Database
4
OECD membership - Czech Republic 1995; Hungary 1996; Poland 1996; Slovakia 2000
5
Estonia and Slovenia joined the OECD only in 2010
4 financial development process by empirically testing its effect on economic growth.
Additionally, capital inflows are also of interest, as growth in V4 may be due to foreign investment, and not innovation. As is revealed in the theory of financial development, better and more developed financial systems lead to higher innovation, as better productivity-enhancing projects are selected and funded. Poorly developed financial
systems will not be able to properly monitor, evaluate, or fund possible innovative projects. The countries of interest in Central Europe are the Visegrad Four (V4), which
consists of the Czech Republic, Hungary, Poland and Slovakia. The remainder of my analysis proceeds as follows: Section 2 of this paper will provide some background and economic trends in Central Europe. In section 3, a literature review of the underlying factors will be outlined, with an emphasis on financial development. Section 4 will discuss the data methods, as well the results of the empirical analysis and a brief discussion. Section 5 concludes with an overall summary of the paper.
2 BACKGROUND
2.1 Transition Economies of Central Europe
The collapse of the former Soviet Union in 1991 gave rise to a number of independent nations. Some of the newly formed countries were prior Soviet states, while others were considered ‘satellite’ states, and include the countries of interest in this paper.
Nonetheless, both groups7 are now generally classified into Central and Eastern European Countries (CEEC). The formation of these new nations subjected their economies to both potential growth and economic instability. Thus, a period of transition arose which saw these economies in a transformation from command economic systems toward market systems. The transition period was required to bring about certain goals, involving the following ingredients (IMF 1999):
7
Another group of nations consists of the Commonwealth Independent States (CIS), but will not be relevant to this paper. They consist of former Soviet Republics.
5 Liberalization the process of allowing most prices to be determined in free markets
and lowering trade barriers that had shut off contact with the price structure of the world's market economies.
Macroeconomic stabilization primarily the process through which inflation is brought under control and lowered over time, after the initial burst of high inflation that follows from liberalization and the release of pent-up demand. This process requires discipline over the government budget and the growth of money and credit (that is, discipline in fiscal and monetary policy) and progress toward sustainable balance of payments.
Restructuring and privatization the processes of creating a viable financial sector and reforming the enterprises in these economies to render them capable of producing goods that could be sold in free markets and of transferring their ownership into private hands.
Legal and institutional reforms these are needed to redefine the role of the state in these economies, establish the rule of law, and introduce appropriate competition policies.
These transitional guidelines were put in place to stabilize the region and pave the way for economic recovery. Additionally, implementing these policies would initiate negotiations and further guidelines for EU membership in the near future. After CEEC fulfilled the required upgrades to institutions and opened their markets to trade, the region would experience a surge of capital flows and rapid economic growth.
2.2 The European Union and Central Europe
The creation of the European Union in 1993 embodied the idea of economic cooperation and growth between its member nations. The introduction of a single currency in 2002 further epitomized the goal of cooperation, having clear transactional cost benefits in relation to trade, while making potential Euro adopting countries adhere to sound
macroeconomic policies. Commitment to these policies is not only beneficial in facilitating the path for adoption of the currency, but it is also instrumental in reducing economic vulnerabilities that would otherwise exist without stern corrections to policies. Coming full circle, adopting sound polices should allow for greater economic cooperation between member nations. It is then complimentary that EU member countries ought to benefit from potential economic growth through both greater cooperation and the resulting enhanced efficiency within the area.
6 Under European Commission guidelines, EU member nations were required to implement full capital account liberation. Such liberalization would then allow the free movement of capital not only between EU member states, but also free capital movement between EU member states and third countries. A schedule of steps8 for capital account liberalization was negotiated between each Central European country and the European Commission. Although each country liberalized at different speeds and under different conditions, three common features were acknowledged (von Hagen et al 2008):
(i) Restrictions on FDI were removed before portfolio flows were liberalized (ii) Capital inflows were liberalized before capital outflows
(iii) Long-term capital flows were liberalized before short-term flows
It is quite implicative that these common features allow for economic growth by
encouraging long-term flows to be properly installed, as well as giving sufficient time for investments to materialize. For example, FDI ensures that an investing nation (firm) has a substantial claim on the returns of an investment. This claim then acts as a monitoring incentive, and will likely prevent the investor from seeking activity that would jeopardize the investment return, such as making risky decisions regarding firm activity. The second feature ensures that capital inflows will persist for a certain period of time, without the possibility of abandoning the resulting investments prematurely. If an investor were allowed to convert his investment into an outflow at any time, this may cause a volatile investment environment that can hinder economic growth, as the necessary investments would not be given the time to develop. The third feature above solidifies the importance of attracting quality, economic-growth driven investments, with key emphasis on the adequate time that such inflows would require to produce a payoff.
8
Von Hagen (2008): All ten countries maintain controls for real estate transactions, and with the exception of Hungary, all have specific provisions with respect to commercial banks and institutional investors.
7 2.3 Recent Economic Trends in Central Europe
2.3.1 Capital Inflows
The transition economies of modern day Central Europe, which includes the V4 sample of interest, began their road to economic growth with one major hurdle: low domestic savings. As detailed in classic macroeconomic models of GDP, investment is an important component of both current and future economic growth. In brief, investment is fuelled by domestic savings, so the higher the savings, the greater the opportunity for investments, and consequentially higher growth. However, as can be expected due to the economic shocks deriving from the collapse of the Soviet Union, domestic savings were low in the initial stages of the transition period in these countries. An alternative and efficient way to fuel growth would be to attract foreign savings in order to induce9 investment. Indeed, since the early 1990’s, large amounts of international private capital flows (in the form of investment) would enter Central Europe in search of higher returns. For example, Austrian private capital flows to Central Europe jumped from more than EUR 1 billion in 1995 to EUR 6.3 billion in 2001, an average annual growth rate of 36% (Ditlbacher et al 2000). When looking at the V4 countries individually from 1993 to 2007, it is evident that the trend of annual capital inflows was pervasive. During this time frame, Czech Republic experienced an increase in FDI inflows10 (in constant US dollars) of $658 million to $10.6 billion; Hungary from $2.3 billion to $70.6 billion; Poland from $1.7 billion to $25.5 billion; Slovakia from $198 million to $3.9 billion. The average increase in FDI inflow growth rates11 were equally impressive, with Czech Republic at 20%, Hungary at 25%, Poland at 20%, and Slovakia at 22%.
To deem capital inflows as the only, if at all, determinant of economic growth in the V4 would be premature. As highlighted in Deuber (2009), Central European countries have relied on a specific growth model based on low savings, high consumption smoothing
9
Caution should also be taken when capital flows into a developing country, inflows of direct
investment were especially prominent in financing external current account imbalances in the Central Europe (Lane and Milesi-Ferretti, 2007).
10
Source: Worldbank.org
11
8 based on potential prosperity gains from EU membership, high dependency on external funding (external savings), and high foreign participation and EU anchor (openness
towards Western Europe). Thus, unilateral transfers from EU to Central Europe, as well as demand for the latter’s goods from the rest of the EU (i.e. economic cooperation) are highly likely to have contributed to their rapid economic growth. These factors will not be taken into account in this paper, but their mention is certainly warranted.
2.3.2 Economic Growth in Central Europe
The transitional period from 1993 to 2007 was characterized by high economic growth rates not only in the V4 group, but also in other 2004 EU accession countries, which together outpaced the growth rates of Western European ones. Figure 1 below, borrowed from von Hagen (2008), is a clear graphical illustration of real growth in the EU from the period 1997-2006.
The clear picture reveals that real growth in the V4 was on average higher than the entire EU area from 1997-2001, but vastly higher on an individual country basis from the period 2002-2007. Deuber (2009) concedes to these findings, exemplifying that Central European
9 average real GDP growth rate was 3.6 percent during 1997-2001, and 5.4 percent during 2002-2006, while the Euro area grew at 2.8 percent and 1.5 percent, respectively.
2.3.3 Financial Crisis in Central Europe
The financial crisis that began to unfold in 2007 is now well documented. The effects were primarily negative in the economic sense and still persistent to this day. Most rich and developed nations in the global economy suffered negative shocks to GDP and
unemployment, while others were even forced to carry potentially unsustainable
government budgets. It is therefore expected that negative spillovers from rich countries would affect the rest of the world. It is estimated that GDP in the US, Euro area, and Japan dropped by 3-4% in 2009, with global trade declining by a staggering 15% (Freund, 2009). At the same time, most of the economies of the V4 and the rest of Central Europe came to a standstill. Additionally, capital inflows in Czech Republic, Poland, and Slovakia decreased modestly during 2008-2009, but substantially decreased in Hungary, and it even
experienced capital outflows. However, the standstill was brief and according to the Global Development Finance 2012: External Debt of Developing Countries, which provides detailed information on the external debt of developing countries, inflows of capital to these countries increased to $1130 billion in 2010, regaining pace at their 2007 pre-crisis peak and highly probable to double the $675 billion recorded in 2009.
The effects of international private capital flows to this region were perhaps instrumental in fuelling GDP growth until the financial crisis began to unwind in 2007. Due to the financial complexities of the crisis, such as banking inter-connectedness between rich-developed nations and those in Central Europe, the economic contractions were evidently present. Central European countries also took a strong hit due to high integration into EU on the trade side and in the banking sector, high share of (export-oriented) industry in GDP and slump in exports, as well as sudden stop in private capital inflows (Deuber, 2009). Thus, it is no surprise that emerging market and developing countries also experienced decreases in GDP. However, central and eastern European countries (CEEC) have
10 rebounded swiftly, and have potentially regained their momentum in GDP growth.
Nonetheless, it is not quite certain how CEEC were affected after the global dip in 2008.
3 THEORY AND LITERATURE REVIEW 3.1 Economic Growth
Economic growth is critical for raising the standards of living for populations around the world. At the very least, if the proper policies for growth are implemented and followed through, the eradication of poverty would be theoretically possible. Research findings have shown that growth is the result of factor accumulation – increases in physical capital and education via investment, increases in labor force participation, and shift from agriculture into manufacturing (Young 1995). The basic model for growth in economics was
developed by Robert Solow in 1956 (see Solow 1956). He uses two simple equations (later expanded in academia to include technology, human capital, education) to depict the production function and capital accumulation.
( ) (1)
(2)
Equation (1) simply states that output (Υ) is a function of capital (Κ) and labor (L). In its most primitive form possible (not accounting for returns to scale and technology),
increasing both capital and labor would have the same effect on output. Equation (2) has a key component very relevant to this paper. It is the capital accumulation equation, which states that the change in capital is the result of investment (in new capital) minus the current period capital depreciation. The first term on the right-hand side, sΥ, is the amount of gross investment. Holding everything else constant, higher investment would lead to higher capital, and ultimately higher output would lead to higher growth. It is then a prevalent assertion that foreign direct investment should lead to economic growth. The Solow model evolves more elegantly, and with the additional work by Mankiw et al (1992), the important concept of conditional convergence in economic growth stood its ground.
11 This simply means that developing nations tend to grow faster than developed ones, with convergence being conditional to lower starting GDP per capita.
For the control variables in my model, it is appropriate to include the findings of the
established research in economic growth. The study by Barro (2003) stems from previous work in the neoclassical growth model such as Ramsey (1928), Solow (1956), and Cass (1965). This article uses cross-country panel regressions to show the differences in per capita growth rates across low-growth countries. The author distinguishes the variables into 2 categories. The first one includes initial levels of state variables, for example, the stock of physical capital available and the stock of human capital in the forms of
educational attainment and health. The second category includes policy variables and national characteristics. Variables in this category comprise of the ratio of government consumption to GDP, international openness, fertility rate, indicators of macroeconomic stability, and rule of law and democracy. In the model, the log of the initial level of per capita GDP enters as an explanatory variable so that the coefficient of the variable represents the rate of convergence, or in other words, the responsiveness of the growth rate (independent variable) to a proportional change in the initial level of per capita GDP. The empirical analysis quantifies a negative coefficient for the log of initial level of per capita GDP, symbolizing the conditional convergence of previous studies. Barro explains that “…the convergence is conditional in that it predicts higher growth in response to lower starting GDP per person only if the other explanatory variables are held constant.” The author asserts that in comparison to the other effects in the model, the value for the coefficient for the log of initial level of per capita GDP has a larger effect, and thus
conditional convergence can have important influences on growth rates. Overall, the paper finds that for a given per capita GDP and human capital, growth depends positively on the rule of law and the investment ratio. It also finds that growth is negatively affected by the fertility rate, the ratio of government consumption to GDP, and the inflation rate. Lastly, growth is found to increase from favorable movements in the nation’s terms of trade and with increased international openness.
12 Prochniak (2011) draws data from the World Bank, IMF, EBRD and the Heritage
Foundation to conduct an empirical analysis of economic growth determinants in Central and Eastern Europe. The determinants are further grouped into (1) demand-side factors, defined as the elements of aggregate demand and exert rather a short-term impact on the rate of economic growth, and (2) supply-side factors, or the inputs to production. To identify the demand-side economic growth determinants, the time series is broken down into three or two-year periods. The author justifies that data averaged over sub-periods identifies demand-side economic growth determinants, but the sub-sub-periods also tend to eliminate short-run fluctuations and long-run tendencies are exposed. The author compiles 47 variables from data, which are classified into the following 11 categories: economic structure, financial sector, households and infrastructure, human capital, international trade, investment, money growth and interest rates, population, private sector, public finance, and other. The first step in the empirical approach is to quantify a coefficient for every variable using partial correlation in order to control for the influence of a third factor, that being the world economic crisis of 2008-2009. Using this basic statistical approach, among the variables with strongest correlations to GDP growth rate include the gross capital formation (% of GDP), Net FDI inflow (% of GDP), Education, Market Capitalization of listed companies (% of GDP), General Government Balance (% of GDP), CPI inflation (%), Population aged 15-64 (% of Total), and technological variables such as Personal Computers (per 100 people) and Mobile phone penetration rate (per 100 people). The paper also finds, after controlling for the impact of the global crisis, a strong negative relationship between GDP growth rate and per capita income from the previous period. This negative relationship means that less developed countries in the group tended to grow faster than developed ones, thus adhering to the existence of unconditional income level convergence. Surprisingly, variables for international trade, such as exports and imports of goods and services, do not display strong relationships with economic growth. As the author points out, however, controlling for the size of the economies in the sample should produce a much stronger relationship. The qualitative variables for structural reforms towards an open economy and changes in the institutional framework also showed strong positive relationships to economic growth. The empirical analysis then evolves to implement OLS regression models, applying the variables with the highest correlations
13 with economic growth. Each model includes GDP per capita from the previous period (to control for the influence of initial conditions) and a global crisis dummy variable. To reduce or eliminate the existence of multicollinearity, variables from the 11 categories are chosen as explanatory determinants of GDP per capita growth. Out of the 10 models selected in his work, 9 of them produced negative and statistically significant coefficients for initial income, once again confirming income level convergence. The regression models concur with the partial correlation analysis, and produce statistically significant
coefficients for the variables mentioned above as determinants of GDP per capita growth.
3.2 Private Capital Inflows
International private capital flows are categorized into three groups, which include foreign direct investment (FDI), portfolio investments, and bank loans. Portfolio investments comprise of purchases of stocks and bonds by foreign agents. This type of investment is mainly driven by portfolio diversification purposes, without the intention to exert control or management influence over a firm. Bank loans refer to foreign lending obtained by either domestic banks or firms. These two types of private capital flows are usually small in terms of nominal value, so they will not be prevalent to the empirical test. However, bank foreign liabilities are captured in another aspect of the empirical test, thus containing an element of bank loans as private capital inflows. As financial markets in the CEEC region continue to develop, it is likely that domestic and foreign investors will begin escalating their investments in equities and debt (stocks and bonds). These types of investments are short-term in nature, and generally are not taken to directly influence firms in the host country. However, increased portfolio investment can be a good indicator for investor sentiment towards this region, as well as a good proxy for financial market development. Foreign bank loan financing provides capital needed for investment when firms are financially constrained. Additionally, increased banking activity, whether led by foreign ownership or domestic institutions, can create further benefits to the economy. Such positive spillovers include credit expansion which leads to further investment, more lending channels within the financial sector which can lead to even further financial
14 integration, and it can signal to foreigners that the host county is perceived as safe for investment as lending requires firms and consumers to meet certain credit standard.
The third type of private capital flows is foreign direct investment (FDI). This type of private capital flow is highlighted in this paper for its potential effects on economic growth. FDI takes place when a firm in one country seeks to exploit foreign market opportunities by expanding production capabilities in those markets. This is can be conducted by either establishing a new firm, merging with or acquiring an existing firm, or cooperating via joint venture with firms in the ‘host’ country. The proper terminology labels the investor as the ‘source’ or ‘source country’, and the term ‘host’ is given to the investee, or recipient
country. Unlike other the other types of private capital flows, the firm engaging in FDI seeks to influence decision making and take control of the investee. A good candidate for attracting FDI usually involves developing economies with sound structural policies and institutions, along with characteristics of comparative advantages such as cheap labor force, as well as low corporate tax rates. The transition economies of the Visegrad Four countries provided such an environment for attracting investment.
FDI can create spillovers in the host country that can further increase growth. In an article by Blomstrom and Kokko (1998), they distinguish three main channels for technology (spillovers) transfers through FDI. The first channel is via competition, where the entrance of foreign firms contributes to development on the industrial, technological, and
managerial levels. The second channel is the ‘demonstration of differences’ in technology between source investors and host firms. In other words, foreign firms are perceived to have better production methods or higher efficiency practices made available by superior technology. Domestic firms then learn by adopting this technology or by replicating these more efficient methods of production. The third channel is via the labor force movement, where local employees gain new productive knowledge from working in these foreign firms. This attained skilled enhancement is then transferred over to the new firm, where potential benefits arise for both the new hiring firm and their current employees.
15 3.3 Financial Development
The importance of financial development for economic growth finds its roots in the classic work of Bagehot (1873). He promoted the critical role of the banking systems in economic growth by prescribing that banks could spur innovation and future growth by identifying and funding productive investments. Almost 40 years later, Schumpeter (1911) further polished the notion that banks play an important role in economic growth by establishing links between the financial and real sectors. He also argued that financial services were an important factor in promoting economic growth. These ideas were formally advanced and expanded in the formative contributions of Goldsmith (1969) and McKinnon (1973) and Shaw (1973), to which the new literature owes much of its foundations.
As made eminent above, economic growth depends on a society’s ability to accumulate savings, subsequently turn these savings into investments, and finally transform these investments into capital for production uses. The capacity to conduct this process requires certain levels of market efficiency to facilitate the flow of resources from one stage to another. Financial development involves the establishment and expansion of institutions, instruments, and markets that support the investment and growth process. A stable and sound system of banks and financial institutions ease society’s burden of holding financial resources for which no adequate use has been identified. The importance of financial intermediation is also relevant here, as it supports the investment process by mobilizing both foreign and domestic savings for firm investment utilization. Levine (1996) describes five main functions which financial systems perform:
facilitate the trading, hedging, diversifying, and pooling of risk allocate resources
monitor managers and exert corporate control mobilize savings
facilitate the exchange of goods and services
A detailed description of the mechanics behind these five functions is not necessary for this paper. It is sufficient to know that these functions of financial systems help promote more transparent and efficient guidelines for channeling resources into future economic growth.
16 Levine further explains how particular market frictions motivate the emergence of
financial markets and intermediaries that provide these five financial functions, and describes how these financial functions affect economic growth via capital accumulation and technological innovation. In another highly cited paper, McKinnon (1973) asserts that the liberalization of financial markets allows for increased activity in financial
intermediation by savers and investors, which ultimately allows the flow of financial resources to be efficiently allocated between investors and banks.
In the research field of financial development, it is almost customary to address the findings of King and Levine (1993a). In their pivotal paper, the authors study whether greater levels of financial development are positively associated with current and future economic growth. They use four indicators to define and measure financial development. The first, financial depth, measures the overall size of the formal financial intermediary system, defined by the ratio of liquid liabilities to GDP. The second distinguishes among financial institutions conducting intermediation. The reasoning is that banks are more likely to provide efficient risk management and investment information services to consumers than central banks. The third and fourth indicators measure where the financial system distributes assets, as funding private firms translates to (potentially) higher provision of services, as opposed to the limited services the financial system could provide to the government or state enterprises. These indicators are captured by both credit issued to nonfinancial private firms divided by total credit (not including credit allocated to banks) and credit issued to nonfinancial private firms to GDP. These fours indicators of financial development are then linked to growth by the rate of physical capital accumulation and the ratio of investment to GDP. Their empirical results confirm that higher levels of financial development have a positive association with physical capital accumulation, economic efficiency improvement, and thus faster rates of economic growth. Additionally, the authors assert that finance does not merely follow economic activity. This assertion is made by their findings in exploring the relationship between future rates of long-run economic growth and the level of financial development. Specifically, they find that financial development is a good predictor of long-run growth over the next 10 to 30 years in their sample of 80 countries over the period 1960-1989.
17 In their follow-up paper, King and Levine (1993b) point to entrepreneurs as having key roles in contributing to economic growth by way of innovation. Moreover, they stipulate the importance of the financial system (financial intermediaries and securities markets), deeming it facilitator of bringing potential entrepreneurial projects to life, by means of expanding the scope and improving the efficiency of innovative activity, and thus accelerating economic growth. They develop an endogenous model that highlights the demand for four services of the financial systems. These services consist of assessing entrepreneurs, accumulating and channeling savings to support the most capable productivity-enhancing projects, diversifying the risks related to these projects, and disclosing the expected profits from participating in innovation. The authors find that financial systems which efficiently appraise, manage, and fund entrepreneurial projects are more developed and lead to higher productivity and economic growth. To empirically test for a linkage between financial systems and growth, they use similar variables as in their work in 1993a. The variables measure financial depth, capture the importance of banks relative to central banks in providing the financial services of their endogenous model, and allocation of credit. Their results once again confirm the positive correlation between financial development and economic growth.
An addition to the earlier works of Levine (1993a and 1993b), pertaining to financial development and economic growth, was produced by Beck et al (2000). They conduct an empirical study to test the relation between the level of financial intermediary
development and several indicators, namely economic growth, physical capital accumulation, total factor productivity, and private savings rates. Highlighting the
differences between this work and the previous ones mentioned are the implementation of more sophisticated econometric techniques. They apply an instrumental variable
estimator to isolate the exogenous component of financial intermediary development, as well as controlling for biases related with simultaneity and unobserved country-specific effects using a GMM (Generalized-Method-of-Moments) panel estimator. This panel estimator also allows for the inclusion of lagged dependent variables as regressors and exploits the time-series variation in the data. They also posit to construct a better measure
18 of financial intermediary, observed as the value of credits by financial intermediaries to the private sector divided by GDP. In contrast to earlier attempts at capturing financial
intermediary, their measure excludes credit issued by both development banks and the central bank, as well as excluding credit to the public sector. Their results confirm earlier findings in the literature regarding the positive relation between financial intermediary development and both real per capita GDP growth and total factor productivity. Although there is an evident positive relation between financial intermediary development and both private savings rate and physical capital accumulation, both indicators of growth are sensitive to robust analyses, producing ambiguous results.
The issue of causality between real GDP and financial development was first studied by Gupta (1984) and later by Jung (1986). However, the works of Demetriades and Hussein (1996) addressed causality with more advanced econometric techniques than their predecessors. More specifically, they examined the issue of non-stationary variables and the integration properties of the data. They first apply cointegration testing before causality testing, as the former has implications for the way the latter is carried out. The authors assert that cointegration testing has an additional advantage of providing
confirmation of a stable long-run relationship between financial and economic
development. They gather data from the IMF publication, International Financial Statistics, and use the usual variables to measure financial development, which includes the ratio of bank deposit liabilities to nominal GDP and the ratio of bank claims on the private sector to nominal GDP. According to the authors, both ratios are valuable in gauging the stage of financial development at a point in time. Their sample includes 16 countries with at least 27 continuous yearly observations for the variables concerned. Although they do not find substantial evidence that economic development follows development in the financial sector, they do find strong evidence of bi-directionality and even some evidence of reverse causation. Finally, the authors point out that their sample countries may have had different institutional characteristics, as well as differences in implementing policies. Hence, due to the nature of their cross-section equations, which combined countries with different experiences with respect to financial development, their findings are very much country specific and thus varying results will be evident within the sample.
19 An advanced empirical study of financial development and economic growth using panel unit root tests and panel cointegration was conducted by Christopoulos and Tsionas (2004). Using a sample of 10 developing countries, these econometric techniques not only account for potential biases induced by simultaneity, country-specific effects, and omitted variables, but also from what the authors call the “threshold effect”, which postulates that under a specific level of financial development there is no effect on economic growth, and a substantial effect above the threshold. A distinction is made between short-run and long-run causality, as most of the gains from financial development may be realized in the short-run, while these same benefits subsequently erode in the long-run as the economy grows. The authors assert that testing for only long-run causality would lead to an inaccurate inference, namely the lack of any causal relationship between output growth and financial development. Their findings confirm that output growth is structurally related to financial development (financial depth variable), and therefore, a long-run relationship is evident. They find no evidence of bi-directionality causality. Further, the authors state that the absence of short-run causality should signal to policy makers that focus should be on long-run policies (e.g., modern financial institutions in the banking sector and stock markets). Additionally, they stress that the long-run nature of the effect on financial development is a necessary implication of the fact that financial markets affect the cost of external finance to the firm, and thus their effect occurs through easing the investment process itself, thus leading to economic growth.
Previous works in stock market development (e.g., Levine (1991), Bencivenga et al (1995)) highlight the aspect of liquidity as a crucial service provided by this type of institution. Moreover, stock markets provide higher liquidity, which induces long-term investments as investors can simply sell their share before the investment matures. However, this
evolution of higher liquidity can also pose potential problems. As liquidity facilitates share selling, it discourages the incentive to monitor managers, which can then lead to
macroeconomic conditions that inhibit effective resource allocation and hamper
productivity growth. Levine and Zervos (1998) thus test whether stock market liquidity, volatility, and size are correlated with current and future rates of economic growth. They
20 measure stock market size as the value of listed domestic shares on domestic exchanges divided by GDP. Stock market liquidity is measured by the value of stock trading relative to the size of the market and by the value of trading relative to the size of the economy. Stock market volatility is measured by a 12-month rolling standard deviation estimate that is based on market returns. These measures are then linked to economic growth by similar channels as previous studies, mainly capital stock growth and productivity growth. Their empirical results confirm that stock market development (as well as development in the banking sector) has a positive and significant correlation with simultaneous and future rates of economic growth, and more specifically, to capital accumulation and productivity growth. The power of their results warrants an explicit example. The authors assert that a one-standard deviation increase in initial stock market liquidity would increase per capita growth by .008% per year over this period (1976-1994). Furthermore, accumulating over 18 years (starting in 1976), this implies that real GDP per capita would have been over 15% higher by 1994. An even more prominent assertion is made when taking both stock market development and banking development into the growth process. Taken together, their findings suggest that if a country had increased both developments (stock market and banking) in 1976 by one standard deviation, then by 1994 real per capita GDP would have been 31% higher, capital stock per person 29% larger, and productivity would have been 24% more efficient.
A study similar and relevant to the theme of this thesis was conducted by Adarov and Tchaidze (2011), which analyses the development of financial markets in Czech Republic, Hungary, Poland, and Slovakia (termed “CE4” countries). They acknowledge a
developmental gap between the size of government bond markets and private bond, private credit, and equity markets, with the latter markets considerably smaller in size. To determine the nature of this gap, they implement variables for institutional development, recognizing their crucial inter-play with other macroeconomic variables, and
acknowledging that positive macroeconomic conditions facilitate institutional
development. To benchmark financial market development in CE4 countries, they compare 68 economies with similar economic development and macroeconomic structural
21 in CE4 countries are significantly less developed given their stage of economic
development as measured by real GDP per capita, and point to the underdevelopment of institutions for this discrepancy. Furthermore, they take account of the effects of external funding on financial market development, as the ability of a home country to borrow from abroad may hinder or stimulate market development. On the negative side, dependence on external funding sources would make financial markets at home less relevant,
under-utilized, and thus lacking the necessary activity for development. Additionally, relying on external funding could be a sign of inability to raise financing at home, and an indication of underdeveloped domestic capital markets. On the positive side, external funding would increase the supply of capital in circulation, leading to stimulate financial market
development. They find that external funding indeed suggests stimulation for financial market development in the CE4 countries. However, in the case of the private bond market, the results suggest that domestic funding is substituted by foreign funding, and thus hindering the development of this particular market.
The issue of whether domestic investment efficiency (and economic growth) is increased by the type of financial structure12 (bank-based or stock-based) a country adheres by remains a normative argument in the literature. Advocates that promote a financial system heavily dependent on banks highlight its ability to issue funds for investments, enhance investment risk by reducing liquidity risk, and reduce the cost of gathering information. Those preferring a stock-based financial system point that stock markets supply better information about the investment’s profitability, reduce the cost of capital via higher liquidity, and exert pressure on corporate management, all of which serve to ease efficiency in resource allocation. An empirical study by Ndikumana (2005) compiles a sample of 99 countries for the period 1965-1997. The author compares both measures of financial development and financial structure to what is referred to an “accelerator-enhancing” effect, which suggests that increases in demand for output (economic growth) is complemented by a rise in the demand for investment. The results reveal that only financial development wields a positive and significant effect on investment via this
12
This thesis includes measures for both bank and stock market systems, thus a brief mention about financial structure is merited
22 “accelerator”, while measures of financial structure have no incremental effect on domestic investment. These results are challenged by Luintel et al (2008), who conclude that financial structure significantly explains output levels in most countries. Their main argument stems from the econometric techniques used in previous studies, asserting that cross-country data cannot be pooled. They use a sample of 14 low and middle-income countries using both dynamic heterogeneous panel and time series methods. The results provide confirmation of significant cross-country heterogeneity in the association between economic growth, financial development, and financial structure.
Although financial development has been linked to economic growth in previous studies, it has also been associated to increase financial fragility. Loayza and Ranciere (2006)
distinguish between short-term and long-term effects of financial development on economic growth. Drawing their ideas from previous theoretical developments, they recognize two alternative views on the consequences of financial development. One view, identified by endogenous growth literature (e.g., Roubini and Sala-i-Martin, 1992), views financial intermediation (financial development) as leading to more efficient allocation of savings for investment opportunities, and thus economic growth. The second view, or the financial crisis literature (e.g., Gavin and Hausman, 1995), highlights the latent
destabilizing effect of financial liberalization as it may lead to an overly expansion of credit. The authors find a negative short-run relationship between financial intermediation and economic growth. To better link this negative relationship, they use both financial volatility, defined as the standard deviation of growth rate of private domestic credit to GDP, and systemic banking crisis, defined as the frequency of years under crisis. They confirm a significant and positive correlation between volatility and crises. They find significant negative correlations between estimated short-run coefficients (for each country) and both volatility and financial crises, leading to the conclusion that higher frequency of crises and higher volatility are related to negative short-run impacts of
financial intermediation on economic growth. Finally, the authors use a growth regression, selecting data from 82 countries between the periods 1960-2000. They use the standard control variables in growth literature, including the initial level of per capita GDP,
23 short-run regressions, they select the variable financial depth to measure financial
development, defining the former as private domestic credit as a ratio to GDP. Again, they include both volatility and crises variables, and find that together or individually, financial banking crises and financial volatility present a significantly negative coefficient. In the long-run growth regression, just as in the short-run regressions, the variable financial depth upholds its significant and positive coefficients.
4. EMPIRICAL ANALYSIS
To determine the relevance of financial development and capital inflows in the Visegrad 4, panel data regressions with fixed effects will be implemented. First, specifications for the country sample and sources for data collection will be introduced. Second, the variables of interest will be defined and the empirical model will be described. Summary statistics are discussed to check for data consistency, followed by description of the empirical results. Third, testing for unit root will give further information about the stationary nature of the variables. Finally, a discussion of the results concludes this section.
4.1 Data
4.1.1 Country Sample
The EU enlargement of 2004 included the following ten countries, which are broken down into two groups to differentiate in this section only: The six, which includes Cyprus,
Estonia, Latvia, Lithuania, Malta, and Slovenia. The other group is the Visegrad Four (V4) and consists of the Czech Republic, Hungary, Poland, and Slovakia. I based my country group selection on GDP, as this is a more telling variable of a country’s contribution to the EU. This is not to say the other six nations are not relevant. To the contrary, they represent a huge significance in geo-political matters and EU stability. However, in 2008, the V4 countries combined for a total GDP (in current $USD) of $1 trillion, compared to a grouped $193 billion from the other six nations13. Of course, both groups can be analyzed together
13
24 in a different theme such as regional growth or financial contagion, but this will not be the focus of this paper.
4.1.2 Data Collection
The majority of the data used for empirics was obtained from the World Bank database website. This database draws data from other sources as well, including the IMF and OECD. For the years 1993-2011, 19 data values were missing from data obtained from the World Bank. Using data from other sources, 17 of the 19 values were established. For example, data for financial openness were obtained from International Financial Statistic manuals published by the IMF; international trade figures for Poland were found on the World Trade Organization website; FDI missing value for Slovakia was supplemented from the National Bank of Slovakia. The two remaining missing observations are for the single category ‘stock market capitalization value’ for the year 1993. Due to country structural reasons, mainly that trading on financial markets did not commence until late in the year 1993, both Czech Republic and Slovakia did not report figures for this category in 1993.
4.1.3 Data Variables
The variable GROWTH represents the real GDP per capita growth rate for the period 1993-2011. Traditional research in GDP per capita growth includes education (Barro 2000), while other theoretical and empirical studies find that growth is positively influenced by the rule of law, capital formation, fertility, terms of trade, human capital, income
distribution, trade, and political stability. On the other hand, high levels of inflation and government consumption are deemed to negate economic growth. The first four variables in the equation above will serve as control variables for the growth regression. They include gross capital formation (% of GDP), education (log of % gross secondary school enrolment), inflation (in % form) and trade openness (the log of ratio of exports plus imports to GDP), and they are labeled CAPTLFORM, LNEDUC, INFLATION, and LNTRADE respectively. All control variables are expected to have positive coefficients, with the exception of inflation, which should have a negative coefficient.
The last four variables are used to measure the relevance of capital flows and
financial development in the V4 countries. Private Capital inflows are directly measured by net inflows of FDI, expressed as a percentage of GDP, and identified by the variable FDI.
25 The World Bank classifies foreign direct investment as the sum of equity capital,
reinvestment of earnings, other long-term capital, and short-term capital used to acquire a management interest the source country. Additionally, portfolio equity inflows were considered for in constructing an aggregate variable for capital inflows. However, this measure was relatively low in terms of GDP, thus predictive power is believed to neither be lost nor enhanced with exclusion. The variable STOCKCAP measures the total value of listed companies registered for trading on each country’s stock markets, and along the lines of Levine and Zervos (1998), serves as an indicator for market development. FINCLOPEN is a variable constructed as the ratio of banks’ aggregate foreign assets and foreign
liabilities to GDP, and measures the degree of world financial integration for the sample group. The variable PRIVCREDIT is the final indicator of financial development, which measures financial depth, and is defined as total credit to private sector in terms of GDP. These four variables are all expressed as a percentage of nominal GDP. The variable FDI is expected to produce a positive and significant coefficient, and thus been influential to economic growth. The other three variables for financial development, according to the literature presented, should have positive and significant coefficients.
4.1.4 Regression Model
The applied model for validation of the variables’ impact on economic growth will be a panel data model with fixed effects. Panel data estimations combine elements of cross-section and time series regressions, and also allow for estimation techniques that account for heterogeneity between variables. According to Baltagi14, panel data models are better suited to study the dynamics of change, give more variability, informative data, degrees of freedom, efficiency, and less collinearity among variables. Fixed effects regression controls for omitted variables in panel data when these omitted variables vary across entities (countries in the context) but do not change over time. Moreover, fixed effects in this model examine the relationship between independent and dependent variables within each country, since each one has its own individual characteristics that may or may not influence the variables. The regression equation below explains the formulation of the empirical model to be tested using panel data statistical methods. The conventional notation in panel
14
26 data modeling is used to represent data for each country. The entity being observed is represented by the subscript i, and the subscript t represents the year of observation:
( ) ( ( )) ( ( )) ( ( )) ( ( )) ( ( )) ( ( )) ( ( )) ( ( )) ( )
4.1.5 Data Summary
Descriptive statistics for the variables are found in Table 1 of the Appendix section. Data values were extracted for a time period of 19 years, pertaining to the four sample countries. A total of 76 observations were established for 7 out of the 8 variables, with the exception being STOCKCAP, which had 74 observations due to missing data in 1993. Table 1 reports the mean real GDP per capita growth rate as 3.46%, with a max value of 10.60% and a min of -6.80%. Inflation displays a max of 36.87%, a min of .11%, and with average inflation rate at 5.50% for the group. This average figure for inflation is rather high, and in addition to the gap between max and min, may cause an ambiguous estimation for its coefficient in the model. The mean value for FDI was 5.51%, but with a standard deviation of 8.43%. This is, in relative terms, the largest reported variance in the group of variables, and with a max value of 51.90% and a min of -16.07%, the figures may also lead to inconsistent coefficient estimation. Mean values for STOCKCAP, FINCLOPEN, and PRIVCREDIT are 17.78%, 15.65%, and 41.80% respectively. All three variables display large gaps between min and max values.
The correlation matrix shows the statistical association between the variables on a scale of -1 to 1, where -1 indicates a perfect negative correlation, and 1 indicating a perfect positive correlation. Values over .5 in absolute terms point to a strong correlation. The correlation values do not give any information in regards to causal effects between the variables. The full results can be found in Table 2 of the Appendix section. According to these results, growth has semi-weak, but negative correlation values with trade openness, financial integration in world markets, and credit to the private sector, with correlations of .23, -.30, and -.35, respectively. Variables for capital formation and credit to the private sector share a correlation of .36, giving some weak support to the idea that financial development
27 increases efficiency in investment allocation. However, the size of the stock market has a negative sign, while financial integration into world markets displays a positive value, but both are strongly-weak. All four variables (non-traditional) of interest share positive correlation values with the measure for trade openness. The correlations are .16 for FDI, .04 for PRIVCREDIT, .65 for STOCKCAP and .40 for FINCLOPEN, with the last two indicating very strong positive correlations. The three variables for financial development produce semi-weak negative correlation values with the measure for inflation. A pre-mature assertion to these correlations regarding inflation is that financial development could have been hampered by volatility in inflation rates witnessed in this group of countries during the early years of the period 1993-2011. Timeline figures depicting growth and key variables can be found in Appendix 2.
4.2 Empirical Results
Four regressions were tested for robustness in the model, in order to observe the sensitivity of the standard growth variables. The four standard growth variables,
CAPTLFORM, LNEDUC, INFLATION, and LNTRADE were included in every regression test.
The first regression tests these four standard variables (as a group) for their effects on growth. Subsequently, the other four variables of interest, FDI, STOCKCAP, FINCLOPEN, and PRIVCREDIT, were individually included in the regressions, ultimately building one cumulative and final regression with all eight variables. For example, FDI enters in the second regression, followed by both FDI and STOCKCAP in the third regression, and so on. The full results are found in columns 1-4 in Table 3 of the Appendix section. The variables for education and capital formation produce positive and significant coefficients in all four regressions. The variable for openness to trade produces four positive coefficients,
although none of the four are significant. Surprisingly, the coefficients for inflation are all positive, including three at the 1% significance level. The statistical results for the main regression model are found in column 5 of Table 3 in the appendix section. All four standard growth variables are positive and statistically significant at the 5% level. Once again, inflation is positive and statistically significant, which contradicts results in the
28 established growth literature. These results for inflation persisted even after introducing 8 lags into the model. This adverse result could signal model misspecifications, or can be attributed to the high levels of inflation experienced during the early years of transition for this group – for which real growth was still present in some of these countries.
The coefficient estimates for the variables of interest are found in column 5 of Table 3 in the Appendix section. The model estimates a negative coefficient of -.001 for FDI, but of no statistical significance. This means that capital inflows, as measured by foreign direct investment into these countries, have had no statistical relevant impact on economic growth. This result strongly contradicts the assertion of this paper that FDI has been a prominent factor for the group’s experience of high economic growth. Additionally, three of the regressions used for variable sensitivity, and which included FDI as a variable, produced both negative and positive coefficients for this variable which were not
significantly different than zero. Only one of the three variables for financial development produced significant figures. The coefficient estimated for FINCLOPEN was -.029, but not statistically significant, meaning that for the group as a whole, financial integration into world markets had no effect economic growth. The measure for financial depth, identified by the variable PRIVCREDIT, produced a coefficient of -.078 at a less than 1% significant level. Thus, the allocation of credit to the private sector has had a negative effect on economic growth during the period 1993-2011. The variable STOCKCAP is accompanied by a positive coefficient of .067, but not significant at the 10% level, although it is
statistically different from zero at the 11% significant level. The importance of the size of the stock markets in this region is therefore rather ambiguous in its effect on economic growth.
4.3 Long-Run Relationships: Unit Root and Cointegration Testing
It is necessary to address the possibility of ‘spurious regression’ when testing for long-run relationships between variables. Spurious regression occurs in a time series when the variables are not stationary. A spurious regression can lead to statistically significant coefficients between variables when in fact these variables of interest are not actually related in practical terms. This false relationship is due to the nonstationarity, or stochastic trend, of the variables in the series. Since time series regressions test observations over
29 time, a basic requirement for proper estimations is that the variables are stationary, with each having a constant mean and variance.
In order to examine if the variables are non-stationary, the basic unit root test will be implemented. This test can be conducted using the Augmented Dickey-Fuller (ADF) test, with the null hypothesis specifying that the variable is non-stationary, and thus considered to have a unit root. The benchmark figure for rejection or non-rejection of the null
hypothesis is the Dickey-Fuller critical value, which is automatically generated in statistical packages. Values are presented at the 1%-5%-10% levels, with the 5% critical value being the most widely used for testing. Rejection of the unit root hypothesis will occur if the estimated test statistic is less than the generated Dickey-Fuller 5% critical value. Otherwise, non-rejection and acceptance of unit root persists until the series becomes stationary. To do so, the ADF model is supplemented with differences of the tested
variable. It is then said to be stationary, or more specifically, integrated with respect to the number of times the variable’s difference was introduced into the ADF test. For example, integrated of order one, also denoted as I(1), would mean that the series was differenced once. Table 4 below presents the order of integration for every variable in each country. The complete table with appropriate T-statistics and critical values is included under Table 4A in the Appendix section.
TABLE 4: Unit Root Testing
Variable Order of Integration per Country
Czech Republic Hungary Poland Slovakia
GROWTH I(1) I(1) I(1) I(0)
CAPTLFORM I(2) I(2) I(3) I(1)
LNEDUC I(2) I(1) I(1) I(1)
INFLATION I(0) I(1) I(0) I(0)
LNTRADE I(1) I(1) I(1) I(1)
FDI I(1) I(1) I(1) I(1)
STOCKCAP I(1) I(1) I(1) I(1)
FINCLOPEN I(2) I(2) I(2) I(1)
PRIVCREDIT I(2) I(2) I(2) I(1)
NOTE: All variables evaluated at the Dickey-Fuller 5% level Critical Value
30 The results of the unit root test show that the majority of the variables become stationary after the first difference, meaning that the majority of stochastic trends are eliminated after the variables’ first differences are applied in the ADF test. Inflation is stationary at the level for all countries except for Hungary. A maximum of three differences appear for the
variable CAPTLFORM. The variable GROWTH is integrated I(0), or at the level, only once. The variables FINCLOPEN and PRIVCREDIT are integrated to the second order in all countries except for Slovakia. Relating this to the empirical model, applying differences to the series will not give any information about the long-run relationships of the variables. The differenced series will only give support to explain the short-run effects of the
variables. Furthermore, nonstationary variables (unit root) are accepted when testing for a long-run relationship, or by examining if the variables are cointegrated.
Cointegration occurs when two or more series have a common stochastic trend. In other words, two nonstationary time series are said to be to be cointegrated if they have a tendency to move together through time. If so, the series exhibits a long-run relationship. To test for cointegration, the Engle-Granger two-step method will be used. This test
assumes that all variables are not stationary, but become stationary when first differenced, or in the order of I(1). The first step is to regress the series using least squares estimation. The second step is to estimate the residuals and once again perform the Augmented-Dickey Fuller test to check for nonstationarity. Although the ADF t-statistic is still the correct computed value for testing the null hypothesis, that the residuals are nonstationary, a different critical value is required. The new MacKinnon critical value takes account of the number of variables included in the test, and can be found on multiple sources15. Rejection of the null hypothesis, or an ADF test statistic less (larger in absolute terms) than the Mackinnon critical value, will result in cointegration among the variables. Since economic research in growth theory has established that the four classic growth variables in this paper are good predictors (positive or negative) of economic growth, and due to the small number of observations per time series (degrees of freedom for a later test), only the four variables (FDI, STOCKCAP, FINCLOPEN, PRIVCREDIT) constructed to test the hypothesis of
15
31 this paper will be included in this two-step examination. The results are presented in Table 5 below.
TABLE 5: EG-ADF Test for Cointegration
Mackinnon Critical Value at 5%: -4.1000
Country ADF Statistic Cointegration
Czech Republic -4.713 YES
Hungary -6.396 YES
Poland -2.719 NO
Slovakia -3.752 NO
Both Czech Republic and Hungary have cointegration within the tested series, while Poland and Slovakia do not. Long-run relationships are established in the first two countries above, but not in the latter two. The EG-ADF test, however, will only reveal if cointegration is present, and will not give explicit information regarding the number of cointegration equations in the series. Several unit root variables in a time series may be present, and therefore, can have more than one cointegrating relationships. With any number of variables N, the time series can have up to N-1 cointegrating relationships. Before
regressing a Vector Error Correction (VEC) model, which estimates the coefficient of long-run relationship, a second test will be performed using the Johansen16 Test for
Cointegration, and will be applied to all countries. The advantage of this test is that it determines the number of cointegration equations per series, as opposed to only testing for the presence of cointegration in the series. Unlike the EG-ADF test, the Johansen test also includes a possible maximum of N-1 cointegrating residuals in the VEC model. In executing the test, two lags will be included, as recommended by the diagnostic selection-order criteria in STATA. Introducing more lags would severely limit the statistical estimation power due to the low number of observations and thus limited degrees of freedom. The results are presented in Table 6 below.
16
For technical details, please see: Johansen, S. (1991). Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models. Econometrica: Journal of the
